Date: Mon, 8 Oct 2012 14:35:41 +0200 From: Fabian Keil <freebsd-listen@fabiankeil.de> To: Andriy Gapon <avg@FreeBSD.org> Cc: fs@FreeBSD.org Subject: Re: ZFS: Deadlock during vnode recycling Message-ID: <20121008143541.537d8490@fabiankeil.de> In-Reply-To: <5071A071.1020800@FreeBSD.org> References: <76CBA055-021F-458D-8978-E9A973D9B783@scsiguy.com> <506EB43B.8050204@FreeBSD.org> <5071A071.1020800@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--Sig_/T+pdA.9Hn4G.E1AymhzME8V Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable Andriy Gapon <avg@FreeBSD.org> wrote: > In fact here is a real patch that I would like to propose: > http://people.freebsd.org/~avg/zfs-getnewvnode_reserve.diff >=20 > The patch incorporates the kib's patch for extending VFS API that helps t= o avoid > entering the vnode inactive/reclaim path from getnewvnode. >=20 > The patch should fix the problem reported in this thread and the problem = from > "panic: _sx_xlock_hard: recursed on non-recursive sx zfsvfs->z_hold_mtx .= .." > thread that runs in parallel. >=20 > Reviews and testing are welcome. I'd probably replace the intentionally empty "else if (zp->z_unlinked)" block in zfs_freebsd_reclaim() with a comment, but other than that the patch looks good to me. While I'm using nullfs on a couple of jails, I never experienced the panic above and can thus only report that I didn't notice any regressions so far since applying the patch yesterday. Fabian --Sig_/T+pdA.9Hn4G.E1AymhzME8V Content-Type: application/pgp-signature; name=signature.asc Content-Disposition: attachment; filename=signature.asc -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.19 (FreeBSD) iEYEARECAAYFAlByyKAACgkQBYqIVf93VJ01ZQCgtVvK7AyF2rgoEvAk0+wpALS/ f/8An1RtXtLDyv83IFRZZ0YX8pCcTq0k =7PFl -----END PGP SIGNATURE----- --Sig_/T+pdA.9Hn4G.E1AymhzME8V--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20121008143541.537d8490>